Skill v1.0.0
currentAutomated scan100/100version: "1.0.0" name: system-testing description: Use when verifying a fully integrated system against its specification — end-to-end functional and non-functional behavior in a production-like environment, system test planning, entry/exit criteria, and gap analysis between integration and acceptance testing. license: MIT metadata: category: functional version: "2.0" tags: system-testing, end-to-end, black-box, production-like, nfr
System Testing
Tests the complete, integrated product as a black box in a production-like environment against requirements — both functional and non-functional. It sits between integration testing (components talk) and acceptance testing (business agrees).
Entry criteria
Integration suite green · build deployed to a dedicated, production-like environment (same topology, versions, config, feature flags) · test data loaded · known defects triaged.
Coverage map (build a test set per row)
| Area | What to verify | Skill | |
|---|---|---|---|
| Functional flows | Every requirement end to end, incl. negative paths | functional-testing | |
| Data integrity | Values consistent across UI, API, DB, reports, exports | database-testing | |
| Interfaces | 3rd-party APIs, webhooks, files, queues | integration-testing, contract-testing | |
| Security | Authn/z, input handling, secrets | security-testing | |
| Performance | Response/throughput at expected + peak load | performance-testing | |
| Reliability | Failover, retry, timeouts, recovery | chaos-resilience-testing | |
| Usability & a11y | Task success, WCAG | usability-testing, accessibility-testing | |
| Compatibility | Browsers, devices, OS, locales | compatibility-testing, localization-testing | |
| Install/upgrade | Fresh install, upgrade, rollback | installation-upgrade-testing | |
| Ops | Logs, metrics, alerts, backups | observability-testing |
Procedure
- Derive test conditions from requirements and from user journeys/state models (
test-design-techniques). - Prioritize by risk = likelihood × impact (
test-strategist). - Execute; log every failure with environment/build/data version.
- Re-run against each fix; run regression on impacted areas (
regression-testing). - Exit: requirement coverage 100% Must / ≥90% Should, no open Sev1/Sev2, NFR targets met, test summary report issued.
Failure-analysis checklist
Reproduce twice · isolate layer (UI/API/DB/network) · check env drift (config, clock, data) · capture logs+trace IDs · classify: product bug / test bug / env issue / requirement gap.
Pitfalls
Testing in a shared, drifting env; mocked dependencies that hide integration failures; no NFR criteria ("must be fast"); skipping data-migration and reporting paths.
Related
integration-testing, acceptance-testing, test-planning-documentation, smoke-sanity-testing